|
0
|
1 /*****************************************************************************
|
|
|
2 genomeCoverageMain.cpp
|
|
|
3
|
|
|
4 (c) 2009 - Aaron Quinlan
|
|
|
5 Hall Laboratory
|
|
|
6 Department of Biochemistry and Molecular Genetics
|
|
|
7 University of Virginia
|
|
|
8 aaronquinlan@gmail.com
|
|
|
9
|
|
|
10 Licenced under the GNU General Public License 2.0 license.
|
|
|
11 ******************************************************************************/
|
|
|
12 #include "genomeCoverageBed.h"
|
|
|
13 #include "version.h"
|
|
|
14
|
|
|
15 using namespace std;
|
|
|
16
|
|
|
17 // define our program name
|
|
|
18 #define PROGRAM_NAME "genomeCoverageBed"
|
|
|
19
|
|
|
20
|
|
|
21 // define our parameter checking macro
|
|
|
22 #define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
|
|
|
23
|
|
|
24 // function declarations
|
|
|
25 void ShowHelp(void);
|
|
|
26
|
|
|
27 int main(int argc, char* argv[]) {
|
|
|
28
|
|
|
29 // our configuration variables
|
|
|
30 bool showHelp = false;
|
|
|
31
|
|
|
32 // input files
|
|
|
33 string bedFile;
|
|
|
34 string genomeFile;
|
|
|
35 int max = INT_MAX;
|
|
|
36 float scale = 1.0;
|
|
|
37
|
|
|
38 bool haveBed = false;
|
|
|
39 bool bamInput = false;
|
|
|
40 bool haveGenome = false;
|
|
|
41 bool startSites = false;
|
|
|
42 bool bedGraph = false;
|
|
|
43 bool bedGraphAll = false;
|
|
|
44 bool eachBase = false;
|
|
|
45 bool eachBaseZeroBased = false;
|
|
|
46 bool obeySplits = false;
|
|
|
47 bool haveScale = false;
|
|
|
48 bool filterByStrand = false;
|
|
|
49 bool only_5p_end = false;
|
|
|
50 bool only_3p_end = false;
|
|
|
51 bool add_gb_track_line = false;
|
|
|
52 string gb_track_opts;
|
|
|
53 string requestedStrand = "X";
|
|
|
54
|
|
|
55 // check to see if we should print out some help
|
|
|
56 if(argc <= 1) showHelp = true;
|
|
|
57
|
|
|
58 for(int i = 1; i < argc; i++) {
|
|
|
59 int parameterLength = (int)strlen(argv[i]);
|
|
|
60
|
|
|
61 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
|
|
|
62 (PARAMETER_CHECK("--help", 5, parameterLength))) {
|
|
|
63 showHelp = true;
|
|
|
64 }
|
|
|
65 }
|
|
|
66
|
|
|
67 if(showHelp) ShowHelp();
|
|
|
68
|
|
|
69 // do some parsing (all of these parameters require 2 strings)
|
|
|
70 for(int i = 1; i < argc; i++) {
|
|
|
71
|
|
|
72 int parameterLength = (int)strlen(argv[i]);
|
|
|
73
|
|
|
74 if(PARAMETER_CHECK("-i", 2, parameterLength)) {
|
|
|
75 if ((i+1) < argc) {
|
|
|
76 haveBed = true;
|
|
|
77 bedFile = argv[i + 1];
|
|
|
78 i++;
|
|
|
79 }
|
|
|
80 }
|
|
|
81 else if(PARAMETER_CHECK("-ibam", 5, parameterLength)) {
|
|
|
82 if ((i+1) < argc) {
|
|
|
83 haveBed = true;
|
|
|
84 bamInput = true;
|
|
|
85 bedFile = argv[i + 1];
|
|
|
86 i++;
|
|
|
87 }
|
|
|
88 }
|
|
|
89 else if(PARAMETER_CHECK("-g", 2, parameterLength)) {
|
|
|
90 if ((i+1) < argc) {
|
|
|
91 haveGenome = true;
|
|
|
92 genomeFile = argv[i + 1];
|
|
|
93 i++;
|
|
|
94 }
|
|
|
95 }
|
|
|
96 else if(PARAMETER_CHECK("-d", 2, parameterLength)) {
|
|
|
97 eachBase = true;
|
|
|
98 }
|
|
|
99 else if(PARAMETER_CHECK("-dz", 3, parameterLength)) {
|
|
|
100 eachBase = true;
|
|
|
101 eachBaseZeroBased = true;
|
|
|
102 }
|
|
|
103 else if(PARAMETER_CHECK("-bg", 3, parameterLength)) {
|
|
|
104 bedGraph = true;
|
|
|
105 }
|
|
|
106 else if(PARAMETER_CHECK("-bga", 4, parameterLength)) {
|
|
|
107 bedGraphAll = true;
|
|
|
108 }
|
|
|
109 else if(PARAMETER_CHECK("-max", 4, parameterLength)) {
|
|
|
110 if ((i+1) < argc) {
|
|
|
111 max = atoi(argv[i + 1]);
|
|
|
112 i++;
|
|
|
113 }
|
|
|
114 }
|
|
|
115 else if(PARAMETER_CHECK("-scale", 6, parameterLength)) {
|
|
|
116 if ((i+1) < argc) {
|
|
|
117 haveScale = true;
|
|
|
118 scale = atof(argv[i + 1]);
|
|
|
119 i++;
|
|
|
120 }
|
|
|
121 }
|
|
|
122 else if(PARAMETER_CHECK("-split", 6, parameterLength)) {
|
|
|
123 obeySplits = true;
|
|
|
124 }
|
|
|
125 else if(PARAMETER_CHECK("-strand", 7, parameterLength)) {
|
|
|
126 if ((i+1) < argc) {
|
|
|
127 filterByStrand = true;
|
|
|
128 requestedStrand = argv[i+1][0];
|
|
|
129 if (!(requestedStrand == "-" || requestedStrand == "+")) {
|
|
|
130 cerr << "*****ERROR: invalid -strand value (" << requestedStrand << "). Allowed options are + or -" << endl;
|
|
|
131 showHelp = true;
|
|
|
132 }
|
|
|
133 i++;
|
|
|
134 }
|
|
|
135 else {
|
|
|
136 cerr << "*****ERROR: -strand options requires a value: + or -" << endl;
|
|
|
137 showHelp = true;
|
|
|
138 }
|
|
|
139 }
|
|
|
140 else if(PARAMETER_CHECK("-3", 2, parameterLength)) {
|
|
|
141 only_3p_end = true;
|
|
|
142 }
|
|
|
143 else if(PARAMETER_CHECK("-5", 2, parameterLength)) {
|
|
|
144 only_5p_end = true;
|
|
|
145 }
|
|
|
146 else if(PARAMETER_CHECK("-trackline", 10, parameterLength)) {
|
|
|
147 add_gb_track_line = true;
|
|
|
148 }
|
|
|
149 else if(PARAMETER_CHECK("-trackopts", 10, parameterLength)) {
|
|
|
150 if ((i+1) < argc) {
|
|
|
151 add_gb_track_line = true;
|
|
|
152 gb_track_opts = argv[i+1];
|
|
|
153 i++;
|
|
|
154 } else {
|
|
|
155 cerr << "*****ERROR: -trackopts options requires a value (UCSC/GB track definition parameters)" << endl;
|
|
|
156 showHelp = true;
|
|
|
157 }
|
|
|
158 }
|
|
|
159 else {
|
|
|
160 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
|
|
|
161 showHelp = true;
|
|
|
162 }
|
|
|
163 }
|
|
|
164
|
|
|
165 // make sure we have both input files
|
|
|
166 if (!haveBed && !haveGenome && !bamInput) {
|
|
|
167 cerr << endl << "*****" << endl << "*****ERROR: Need both a BED (-i) and a genome (-g) file. " << endl << "*****" << endl;
|
|
|
168 showHelp = true;
|
|
|
169 }
|
|
|
170 if (bedGraph && eachBase) {
|
|
|
171 cerr << endl << "*****" << endl << "*****ERROR: Use -d/-dz or -bg, not both" << endl << "*****" << endl;
|
|
|
172 showHelp = true;
|
|
|
173 }
|
|
|
174 if (bedGraphAll && eachBase) {
|
|
|
175 cerr << endl << "*****" << endl << "*****ERROR: Use -d/-dz or -bga, not both" << endl << "*****" << endl;
|
|
|
176 showHelp = true;
|
|
|
177 }
|
|
|
178
|
|
|
179 if (only_3p_end && only_5p_end) {
|
|
|
180 cerr << endl << "*****" << endl << "*****ERROR: Use -3 or -5, not both " << endl << "*****" << endl;
|
|
|
181 showHelp = true;
|
|
|
182 }
|
|
|
183
|
|
|
184 if ( (only_3p_end||only_5p_end) && obeySplits) {
|
|
|
185 cerr << endl << "*****" << endl << "*****ERROR: Use -split can't be used with -3 or -5." << endl << "*****" << endl;
|
|
|
186 showHelp = true;
|
|
|
187 }
|
|
|
188
|
|
|
189 if (add_gb_track_line && !(bedGraph||bedGraphAll)) {
|
|
|
190 cerr << endl << "*****" << endl << "*****ERROR: Using -trackline requires bedGraph output (use -bg or -bga)." << endl << "*****" << endl;
|
|
|
191 showHelp = true;
|
|
|
192 }
|
|
|
193
|
|
|
194 if (haveScale && !(bedGraph||bedGraphAll||eachBase)) {
|
|
|
195 cerr << endl << "*****" << endl << "*****ERROR: Using -scale requires bedGraph output (use -bg or -bga) or per base depth (-d)." << endl << "*****" << endl;
|
|
|
196 showHelp = true;
|
|
|
197 }
|
|
|
198
|
|
|
199 if (!showHelp) {
|
|
|
200 BedGenomeCoverage *bc = new BedGenomeCoverage(bedFile, genomeFile, eachBase,
|
|
|
201 startSites, bedGraph, bedGraphAll,
|
|
|
202 max, scale, bamInput, obeySplits,
|
|
|
203 filterByStrand, requestedStrand,
|
|
|
204 only_5p_end, only_3p_end,
|
|
|
205 eachBaseZeroBased,
|
|
|
206 add_gb_track_line, gb_track_opts);
|
|
|
207 delete bc;
|
|
|
208
|
|
|
209 return 0;
|
|
|
210 }
|
|
|
211 else {
|
|
|
212 ShowHelp();
|
|
|
213 }
|
|
|
214 }
|
|
|
215
|
|
|
216 void ShowHelp(void) {
|
|
|
217
|
|
|
218 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
|
|
|
219
|
|
|
220 cerr << "Authors: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
|
|
|
221 cerr << " Assaf Gordon, CSHL" << endl << endl;
|
|
|
222
|
|
|
223 cerr << "Summary: Compute the coverage of a feature file among a genome." << endl << endl;
|
|
|
224
|
|
|
225 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome>" << endl << endl;
|
|
|
226
|
|
|
227 cerr << "Options: " << endl;
|
|
|
228
|
|
|
229 cerr << "\t-ibam\t\t" << "The input file is in BAM format." << endl;
|
|
|
230 cerr << "\t\t\tNote: BAM _must_ be sorted by position" << endl << endl;
|
|
|
231
|
|
|
232 cerr << "\t-d\t\t" << "Report the depth at each genome position (with one-based coordinates)." << endl;
|
|
|
233 cerr << "\t\t\tDefault behavior is to report a histogram." << endl << endl;
|
|
|
234
|
|
|
235 cerr << "\t-dz\t\t" << "Report the depth at each genome position (with zero-based coordinates)." << endl;
|
|
|
236 cerr << "\t\t\tReports only non-zero positions." << endl;
|
|
|
237 cerr << "\t\t\tDefault behavior is to report a histogram." << endl << endl;
|
|
|
238
|
|
|
239 cerr << "\t-bg\t\t" << "Report depth in BedGraph format. For details, see:" << endl;
|
|
|
240 cerr << "\t\t\tgenome.ucsc.edu/goldenPath/help/bedgraph.html" << endl << endl;
|
|
|
241
|
|
|
242 cerr << "\t-bga\t\t" << "Report depth in BedGraph format, as above (-bg)." << endl;
|
|
|
243 cerr << "\t\t\tHowever with this option, regions with zero " << endl;
|
|
|
244 cerr << "\t\t\tcoverage are also reported. This allows one to" << endl;
|
|
|
245 cerr << "\t\t\tquickly extract all regions of a genome with 0 " << endl;
|
|
|
246 cerr << "\t\t\tcoverage by applying: \"grep -w 0$\" to the output." << endl << endl;
|
|
|
247
|
|
|
248 cerr << "\t-split\t\t" << "Treat \"split\" BAM or BED12 entries as distinct BED intervals." << endl;
|
|
|
249 cerr << "\t\t\twhen computing coverage." << endl;
|
|
|
250 cerr << "\t\t\tFor BAM files, this uses the CIGAR \"N\" and \"D\" operations " << endl;
|
|
|
251 cerr << "\t\t\tto infer the blocks for computing coverage." << endl;
|
|
|
252 cerr << "\t\t\tFor BED12 files, this uses the BlockCount, BlockStarts, and BlockEnds" << endl;
|
|
|
253 cerr << "\t\t\tfields (i.e., columns 10,11,12)." << endl << endl;
|
|
|
254
|
|
|
255 cerr << "\t-strand\t\t" << "Calculate coverage of intervals from a specific strand." << endl;
|
|
|
256 cerr << "\t\t\tWith BED files, requires at least 6 columns (strand is column 6). " << endl;
|
|
|
257 cerr << "\t\t\t- (STRING): can be + or -" << endl << endl;
|
|
|
258
|
|
|
259 cerr << "\t-5\t\t" << "Calculate coverage of 5\" positions (instead of entire interval)." << endl << endl;
|
|
|
260
|
|
|
261 cerr << "\t-3\t\t" << "Calculate coverage of 3\" positions (instead of entire interval)." << endl << endl;
|
|
|
262
|
|
|
263 cerr << "\t-max\t\t" << "Combine all positions with a depth >= max into" << endl;
|
|
|
264 cerr << "\t\t\ta single bin in the histogram. Irrelevant" << endl;
|
|
|
265 cerr << "\t\t\tfor -d and -bedGraph" << endl;
|
|
|
266 cerr << "\t\t\t- (INTEGER)" << endl << endl;
|
|
|
267
|
|
|
268 cerr << "\t-scale\t\t" << "Scale the coverage by a constant factor." << endl;
|
|
|
269 cerr << "\t\t\tEach coverage value is multiplied by this factor before being reported." << endl;
|
|
|
270 cerr << "\t\t\tUseful for normalizing coverage by, e.g., reads per million (RPM)." << endl;
|
|
|
271 cerr << "\t\t\t- Default is 1.0; i.e., unscaled." << endl;
|
|
|
272 cerr << "\t\t\t- (FLOAT)" << endl << endl;
|
|
|
273
|
|
|
274 cerr << "\t-trackline\t" << "Adds a UCSC/Genome-Browser track line definition in the first line of the output." << endl;
|
|
|
275 cerr <<"\t\t\t- See here for more details about track line definition:" << endl;
|
|
|
276 cerr <<"\t\t\t http://genome.ucsc.edu/goldenPath/help/bedgraph.html" << endl;
|
|
|
277 cerr <<"\t\t\t- NOTE: When adding a trackline definition, the output BedGraph can be easily" << endl;
|
|
|
278 cerr <<"\t\t\t uploaded to the Genome Browser as a custom track," << endl;
|
|
|
279 cerr <<"\t\t\t BUT CAN NOT be converted into a BigWig file (w/o removing the first line)." << endl << endl;
|
|
|
280
|
|
|
281 cerr << "\t-trackopts\t"<<"Writes additional track line definition parameters in the first line." << endl;
|
|
|
282 cerr <<"\t\t\t- Example:" << endl;
|
|
|
283 cerr <<"\t\t\t -trackopts 'name=\"My Track\" visibility=2 color=255,30,30'" << endl;
|
|
|
284 cerr <<"\t\t\t Note the use of single-quotes if you have spaces in your parameters." << endl;
|
|
|
285 cerr <<"\t\t\t- (TEXT)" << endl << endl;
|
|
|
286
|
|
|
287 cerr << "Notes: " << endl;
|
|
|
288 cerr << "\t(1) The genome file should tab delimited and structured as follows:" << endl;
|
|
|
289 cerr << "\t <chromName><TAB><chromSize>" << endl << endl;
|
|
|
290 cerr << "\tFor example, Human (hg19):" << endl;
|
|
|
291 cerr << "\tchr1\t249250621" << endl;
|
|
|
292 cerr << "\tchr2\t243199373" << endl;
|
|
|
293 cerr << "\t..." << endl;
|
|
|
294 cerr << "\tchr18_gl000207_random\t4262" << endl << endl;
|
|
|
295
|
|
|
296 cerr << "\t(2) The input BED (-i) file must be grouped by chromosome." << endl;
|
|
|
297 cerr << "\t A simple \"sort -k 1,1 <BED> > <BED>.sorted\" will suffice."<< endl << endl;
|
|
|
298
|
|
|
299 cerr << "\t(3) The input BAM (-ibam) file must be sorted by position." << endl;
|
|
|
300 cerr << "\t A \"samtools sort <BAM>\" should suffice."<< endl << endl;
|
|
|
301
|
|
|
302 cerr << "Tips: " << endl;
|
|
|
303 cerr << "\tOne can use the UCSC Genome Browser's MySQL database to extract" << endl;
|
|
|
304 cerr << "\tchromosome sizes. For example, H. sapiens:" << endl << endl;
|
|
|
305 cerr << "\tmysql --user=genome --host=genome-mysql.cse.ucsc.edu -A -e \\" << endl;
|
|
|
306 cerr << "\t\"select chrom, size from hg19.chromInfo\" > hg19.genome" << endl << endl;
|
|
|
307
|
|
|
308
|
|
|
309 // end the program here
|
|
|
310 exit(1);
|
|
|
311 }
|
|
|
312
|
